8:39 p.m.
Hi,
I get the following error when trying to build xbill under devel.
I have no problem under F10 and F9.
Bye,
Andrea.
---------- Forwarded message ----------
From: <rpmfusion-buildsys(a)lists.rpmfusion.org>
Date: Thu, Feb 26, 2009 at 8:51 PM
Subject: Build Error (Job 2919): xbill-2_1-2_fc11 on
fedora-development-rpmfusion_free
To: musuruan(a)gmail.com
Job failed on arch ppc64
Build logs may be found at
http://buildsys.rpmfusion.org/logs/fedora-development-rpmfusion_free/2919...
-------------------------------------------------
tcp_wrappers-libs ppc64 7.6-53.fc10 fedora-everything 67 k
tzdata noarch 2009a-1.fc11 fedora-everything 769 k
udev ppc64 137-4.fc11 fedora-everything 338 k
upstart ppc64 0.3.9-19.fc10 fedora-everything 280 k
ustr ppc64 1.0.4-7.fc10 fedora-everything 99 k
zlib ppc64 1.2.3-19.fc11 fedora-everything 79 k
Transaction Summary
================================================================================
Install 126 Package(s)
Update 0 Package(s)
Remove 0 Package(s)
Total download size: 130 M
Downloading Packages:
--------------------------------------------------------------------------------
Total 191 MB/s | 130 MB 00:00
Running rpm_check_debug
ERROR with rpm_check_debug vs depsolve:
rpmlib(FileDigests) is needed by popt-1.13-5.fc11.ppc64
rpmlib(FileDigests) is needed by kernel-headers-2.6.29-0.148.rc6.fc11.ppc64
rpmlib(FileDigests) is needed by procps-3.2.7-27.fc11.ppc64
rpmlib(FileDigests) is needed by bash-4.0-1.fc11.ppc64
rpmlib(FileDigests) is needed by ethtool-6-3.20090115git.fc11.ppc64
rpmlib(FileDigests) is needed by audit-libs-1.7.12-1.fc11.ppc64
Complete!
Cleaning up...
Executing /usr/sbin/mock-helper umount
/var/lib/mock/fedora-development-ppc64-rpmfusion_free-63770729c8223d26481b4aa0edc2f467385b1097/root/proc
Executing /usr/sbin/mock-helper umount
/var/lib/mock/fedora-development-ppc64-rpmfusion_free-63770729c8223d26481b4aa0edc2f467385b1097/root/dev/pts
Done.
9:04 p.m.
On Thu, Feb 26, 2009 at 09:39:24PM +0100, Andrea Musuruane wrote:
I get the following error when trying to build xbill under
devel.
I have no problem under F10 and F9.
I do not know for sure, but with the mass rebuild going on, my mirror has
a hard time keeping up with all the changes and therefore I believe it
could be that some of the mirrors the builders are using are not 100% in
sync. I would say just retry in a few hours again.
Adrian
9:51 p.m.
Adrian Reber wrote:
I do not know for sure, but with the mass rebuild going on, my mirror
has
a hard time keeping up with all the changes and therefore I believe it
could be that some of the mirrors the builders are using are not 100% in
sync. I would say just retry in a few hours again.
No, the problem is that you need to use RPM 4.6 on the host system to build
Rawhide packages, see the discussions on the fedora-devel-list.
Kevin Kofler
9:55 a.m.
On Thu, Feb 26, 2009 at 10:51 PM, Kevin Kofler <kevin.kofler(a)chello.at> wrote:
No, the problem is that you need to use RPM 4.6 on the host system to
build
Rawhide packages, see the discussions on the fedora-devel-list.
If the problem is that our build system doesn't yet use RPM 4.6, why
don't we update it?
Thanks.
Bye,
Andrea.
1:03 p.m.
On 27.02.2009 10:55, Andrea Musuruane wrote:
On Thu, Feb 26, 2009 at 10:51 PM, Kevin Kofler
<kevin.kofler(a)chello.at> wrote:
> No, the problem is that you need to use RPM 4.6 on the host system to build
> Rawhide packages, see the discussions on the fedora-devel-list.
If the problem is that our build system doesn't yet use RPM 4.6, why
don't we update it?
One reason: Because the PPC builder uses F9 and is used for other things
as well -- so simply updating it to a RPM version might break other
crucial services on that machine. Thus is needs to be properly discussed
and planed how to move forward before we can do something. I tried to
reach the owner of the PPC machine (who's traveling right now) yesterday
to discuss a solution, but didn't get a response yet.
Another reason: Somebody needs to sit down and actually do install the
new rpm (a sane one; not one that is found unsigned somewhere on the
net!) on the two x86-builders that both run CentOS 5.
But the real reason is a different one:
*Our organization in regards to infrastructure maintenance sucks. A lot.*
We afaics need a real infrastructure team with at least two or three
*active* (e.g. not only driven by events like breakage) people that take
care of the builders and all the other infra we have. Only then RPM
Fusion will last. I had hoped Xavier would form such a team, but that
didn't happen afaics and he afaics seems to be quite busy with real life
and Fedora-tasks right now, otherwise I guess he would have foreseen
this problem and acted already.
CU
knurd
P.S.: Please note that I don't want to get counted as part of the
"infrastructure team". Yes, I have access to some (nearly all) of the
builders (sometimes I'm the only one that has access), but I really want
to hand the responsibility over to a infra team. But there is nobody I
can hand it over to afaics. Reminder: I just did infra work in the past
because somebody had to do it to get RPM Fusion running.
And another reminder: Our infra is quite fragile afaics. We have lots of
single-points of failure and there are lots of hacks here and there to
make things work. That could become a real problem in the long run if we
don't form a real active infra team.
2:55 a.m.
Thorsten Leemhuis wrote:
One reason: Because the PPC builder uses F9 and is used for other
things
as well -- so simply updating it to a RPM version might break other
crucial services on that machine. Thus is needs to be properly discussed
and planed how to move forward before we can do something. I tried to
reach the owner of the PPC machine (who's traveling right now) yesterday
to discuss a solution, but didn't get a response yet.
Another reason: Somebody needs to sit down and actually do install the
new rpm (a sane one; not one that is found unsigned somewhere on the
net!) on the two x86-builders that both run CentOS 5.
Mitr has built RPM 4.6 packages for F9 and EL5:
http://people.redhat.com/mitr/sha256-rpm/
Several people have upgraded their RPM to these builds successfully, I
haven't read about any breakage yet.
Of course you could also build your own if you don't trust the way these are
published...
Note that if the machines are running stuff like GDB, that stuff also needs
to be rebuilt against the new librpm (or a compat package used, but using a
compat librpm when the system uses a newer RPM might be a source of
trouble).
Kevin Kofler
9:52 p.m.
Thorsten Leemhuis pisze:
On 27.02.2009 10:55, Andrea Musuruane wrote:
> On Thu, Feb 26, 2009 at 10:51 PM, Kevin Kofler
> <kevin.kofler(a)chello.at> wrote:
>> No, the problem is that you need to use RPM 4.6 on the host system to
>> build
>> Rawhide packages, see the discussions on the fedora-devel-list.
> If the problem is that our build system doesn't yet use RPM 4.6, why
> don't we update it?
One reason: Because the PPC builder uses F9 and is used for other things
as well -- so simply updating it to a RPM version might break other
crucial services on that machine. Thus is needs to be properly discussed
and planed how to move forward before we can do something. I tried to
reach the owner of the PPC machine (who's traveling right now) yesterday
to discuss a solution, but didn't get a response yet.
Another reason: Somebody needs to sit down and actually do install the
new rpm (a sane one; not one that is found unsigned somewhere on the
net!) on the two x86-builders that both run CentOS 5.
But the real reason is a different one:
*Our organization in regards to infrastructure maintenance sucks. A lot.*
We afaics need a real infrastructure team with at least two or three
*active* (e.g. not only driven by events like breakage) people that take
care of the builders and all the other infra we have. Only then RPM
Fusion will last. I had hoped Xavier would form such a team, but that
didn't happen afaics and he afaics seems to be quite busy with real life
and Fedora-tasks right now, otherwise I guess he would have foreseen
this problem and acted already.
CU
knurd
P.S.: Please note that I don't want to get counted as part of the
"infrastructure team". Yes, I have access to some (nearly all) of the
builders (sometimes I'm the only one that has access), but I really want
to hand the responsibility over to a infra team. But there is nobody I
can hand it over to afaics. Reminder: I just did infra work in the past
because somebody had to do it to get RPM Fusion running.
And another reminder: Our infra is quite fragile afaics. We have lots of
single-points of failure and there are lots of hacks here and there to
make things work. That could become a real problem in the long run if we
don't form a real active infra team.
I wonder how redhat folks have solved this problem, since afaik the koji
builders are running rhel.
Julian
10:24 p.m.
2009/3/1 Julian Sikorski <belegdol(a)gmail.com>:
Thorsten Leemhuis pisze:
> On 27.02.2009 10:55, Andrea Musuruane wrote:
>> On Thu, Feb 26, 2009 at 10:51 PM, Kevin Kofler
>> <kevin.kofler(a)chello.at> wrote:
>>> No, the problem is that you need to use RPM 4.6 on the host system to
>>> build
>>> Rawhide packages, see the discussions on the fedora-devel-list.
>> If the problem is that our build system doesn't yet use RPM 4.6, why
>> don't we update it?
>
> One reason: Because the PPC builder uses F9 and is used for other things
> as well -- so simply updating it to a RPM version might break other
> crucial services on that machine. Thus is needs to be properly discussed
> and planed how to move forward before we can do something. I tried to
> reach the owner of the PPC machine (who's traveling right now) yesterday
> to discuss a solution, but didn't get a response yet.
>
> Another reason: Somebody needs to sit down and actually do install the
> new rpm (a sane one; not one that is found unsigned somewhere on the
> net!) on the two x86-builders that both run CentOS 5.
>
>
> But the real reason is a different one:
>
> *Our organization in regards to infrastructure maintenance sucks. A lot.*
>
> We afaics need a real infrastructure team with at least two or three
> *active* (e.g. not only driven by events like breakage) people that take
> care of the builders and all the other infra we have. Only then RPM
> Fusion will last. I had hoped Xavier would form such a team, but that
> didn't happen afaics and he afaics seems to be quite busy with real life
> and Fedora-tasks right now, otherwise I guess he would have foreseen
> this problem and acted already.
>
> CU
> knurd
>
> P.S.: Please note that I don't want to get counted as part of the
> "infrastructure team". Yes, I have access to some (nearly all) of the
> builders (sometimes I'm the only one that has access), but I really want
> to hand the responsibility over to a infra team. But there is nobody I
> can hand it over to afaics. Reminder: I just did infra work in the past
> because somebody had to do it to get RPM Fusion running.
>
> And another reminder: Our infra is quite fragile afaics. We have lots of
> single-points of failure and there are lots of hacks here and there to
> make things work. That could become a real problem in the long run if we
> don't form a real active infra team.
>
I wonder how redhat folks have solved this problem, since afaik the koji
builders are running rhel.
rhel 5.3, IIRC which doesn't have the matching
equivalent in centos yet.
Isn't it ?
Nicolas (kwizart)
10:49 p.m.
Julian Sikorski wrote:
I wonder how redhat folks have solved this problem, since afaik the
koji
builders are running rhel.
They upgraded RPM on their build hosts to the 4.6.0 packages mitr built:
http://people.redhat.com/mitr/sha256-rpm/
Kevin Kofler
10:55 p.m.
Nicolas Chauvet wrote:
rhel 5.3, IIRC which doesn't have the matching equivalent in
centos yet.
Isn't it ?
RHEL 5.3 only provides the latest yum which is needed to process the SHA256
metadata. That one is also in centos-testing for CentOS 5.2. That plus
python-hashlib which Seth Vidal built:
http://skvidal.fedorapeople.org/hashlib/ is needed for the metadata, it
doesn't help with the RPM packages themselves using SHA256 checksums
though. You need RPM 4.6 for that:
http://people.redhat.com/mitr/sha256-rpm/
Kevin Kofler
4:26 p.m.
Kevin Kofler pisze:
Nicolas Chauvet wrote:
> rhel 5.3, IIRC which doesn't have the matching equivalent in centos yet.
> Isn't it ?
RHEL 5.3 only provides the latest yum which is needed to process the SHA256
metadata. That one is also in centos-testing for CentOS 5.2. That plus
python-hashlib which Seth Vidal built:
http://skvidal.fedorapeople.org/hashlib/ is needed for the metadata, it
doesn't help with the RPM packages themselves using SHA256 checksums
though. You need RPM 4.6 for that:
http://people.redhat.com/mitr/sha256-rpm/
Kevin Kofler
I asked in #fedora-admin and they said that
http://people.redhat.com/mitr/sha256-rpm/rpm-4.6.0-4.0.mitr.1.el5.src.rpm
was used, and that python-hashlib from epel-testing is needed as well. I
was also told that the above rpm replaces the host's rpm, and that there
is no downgrade possible.
Julian
5:35 p.m.
On 02.03.2009 17:26, Julian Sikorski wrote:
Kevin Kofler pisze:
> Nicolas Chauvet wrote:
>> rhel 5.3, IIRC which doesn't have the matching equivalent in centos yet.
>> Isn't it ?
> RHEL 5.3 only provides the latest yum which is needed to process the SHA256
> metadata. That one is also in centos-testing for CentOS 5.2. That plus
> python-hashlib which Seth Vidal built:
> http://skvidal.fedorapeople.org/hashlib/ is needed for the metadata, it
> doesn't help with the RPM packages themselves using SHA256 checksums
> though. You need RPM 4.6 for that:
> http://people.redhat.com/mitr/sha256-rpm/
I asked in #fedora-admin and they said that
http://people.redhat.com/mitr/sha256-rpm/rpm-4.6.0-4.0.mitr.1.el5.src.rpm
was used, and that python-hashlib from epel-testing is needed as well. I
was also told that the above rpm replaces the host's rpm, and that there
is no downgrade possible.
The new RPM never was the a real problem -- the necessary steps were
well known (I should have made that more clear in my mail, as that could
have avoided lot of others mails in this thread; sorry). Xavier is
working on it and he has access to the two x86 builders now)
The PPC build is the problem and will likely stay a problem for a few
days (¹), as I couldn't reach dwmw2 yet.
CU
knurd
(¹) partly because I am busy over the next few days; if anyone want's to
poke him feel free to do so
2:07 p.m.
Quick status update:
On 02.03.2009 18:35, Thorsten Leemhuis wrote:
On 02.03.2009 17:26, Julian Sikorski wrote:
The new RPM never was the a real problem -- the necessary steps were
well known (I should have made that more clear in my mail, as that could
have avoided lot of others mails in this thread; sorry). Xavier is
working on it and he has access to the two x86 builders now)
Xavier worked on this, but he didn't finish it yet according to what I
read on IRC. He afaics plans to work further on it this weekend.
The PPC build is the problem and will likely stay a problem for a few
days (¹), as I couldn't reach dwmw2 yet.
Haven't reached him yet via mail or IRC (in fact he's not on IRC since a
few days at all, which is odd). Looks like he's on vacation or something.
CU
knurd
5529
days inactive
5538
days old
rpmfusion-developers@lists.rpmfusion.org
12 comments
6 participants
participants (6)
-
Adrian Reber -
Andrea Musuruane -
Julian Sikorski -
Kevin Kofler -
Nicolas Chauvet -
Thorsten Leemhuis